NVIDIA Pledges $250 Million in Computing for UK AI Safety Research

NVIDIA plans to provide computing resources valued at $250 million over five years for AI safety research in the United Kingdom, while also planning investments in five British startups. The computing commitment averages about $50 million annually and represents resources, not a cash grant.

UK cloud partners will provide the computing. Their identities and commercial terms are still being finalized. The research will involve BOLD, led by Oxford University; SOFAIR, led by University College London (UCL); and Imperial College. Areas of study include AI agent safety, open-model security and defenses against real-world threats.

The planned startup investments span several fields. Boltz develops biomolecular AI models; Foresight works on infrastructure risk prediction; Hiverge builds an AI engine for algorithm optimization; Prima Mente develops biological foundation models focused on neurodegenerative diseases; and Riverlane works on quantum-error-correction hardware and software.